Firms Partner on AI Safety
Gen and Vercel are partnering to bring independent safety verification to the AI skills market. Gen's Agent Trust Hub will provide risk verification for skills.sh, a platform for AI agents. The initiative aims to protect developers and users from unsafe AI skills as agents become more autonomous.
- Gen is a global digital safety company with brands like Norton, Avast, and LifeLock, serving almost 500 million users in over 150 countries. - Vercel, valued at over $3 billion, is a cloud platform that created the popular Next.js web development framework and serves more than 6 million developers. - The Agent Trust Hub will rate each AI skill with one of four classifications: Safe, Low Risk, High Risk, or Critical Risk, based on an analysis of potential security flaws or malicious intent. - This verification is crucial as AI agents evolve from simple response generators to autonomous actors that can browse the web, connect to APIs, and access sensitive data, increasing the risk of vulnerabilities and fraud. - The skills.sh platform functions as an open directory, similar to a package manager, where developers can publish and install reusable, modular capabilities to extend what AI agents can do. - Gen's own research from its Threat Labs found that nearly 15% of analyzed skills for the OpenClaw AI agent contained malicious instructions, and over 18,000 instances were exposed to online attacks. - The collaboration aims to embed security directly into the AI development lifecycle, making it a proactive measure rather than an afterthought. - This initiative by Gen and Vercel is part of a broader effort to create a trust infrastructure for what they term the "agentic era," where autonomous AI systems become more widespread.
